Rett syndrome is a neurodevelopmental disorder caused by mutations in methyl-CpG-binding protein 2 (MECP2), a transcriptional regulator. In addition to cognitive, communication, and motor problems, affected individuals have abnormalities in autonomic function and respiratory control that may contribute to premature lethality. Mice lacking Mecp2 die early and recapitulate the autonomic and respiratory phenotypes seen in humans. The association of autonomic and respiratory deficits with premature death suggests that Mecp2 is critical within autonomic and respiratory control centers for survival. To test this, we compared the autonomic and respiratory phenotypes of mice with a null allele of Mecp2 to mice with Mecp2 removed from their brainstem and spinal cord. We found that MeCP2 is necessary within the brainstem and spinal cord for normal lifespan, normal control of heart rate, and respiratory response to hypoxia. Restoration of MeCP2 in a subset of the cells in this same region is sufficient to rescue abnormal heart rate and abnormal respiratory response to hypoxia. Furthermore, restoring MeCP2 function in neural centers critical for autonomic and respiratory function alleviates the lethality associated with loss of MeCP2 function, supporting the notion of targeted therapy toward treating Rett syndrome.
Mutations in the X-linked transcriptional regulator methyl-CpG-binding protein 2 (MECP2) are common in individuals with Rett syndrome (RTT, MIM 312750) (Amir et al., 1999). This disorder predominantly affects girls and is characterized by loss of spoken language and loss of hand skills (Neul et al., 2010). A critical concern in RTT is the incidence of sudden unexpected death, which causes 25% of all deaths (Kerr et al., 1997). Although the exact cause of such deaths is unknown, breathing abnormalities (Julu et al., 2001; Weese-Mayer et al., 2006; Rohdin et al., 2007) and cardiac rhythm irregularities (Guideri et al., 1999; Madan et al., 2004) are suspected to be causative. Characterization of breathing and heart rate revealed a lack of coordination between the two, suggesting alteration in the medullary network that integrates these systems (Julu et al., 2001). Most boys with MECP2 mutations are severely affected with congenital encephalopathy and death within 3 years of life associated with respiratory arrest (Kankirawatana et al., 2006) and marked bradycardia (Schüle et al., 2008) at death.
Mice lacking MeCP2 function reproduce features of the human disorder (Chen et al., 2001; Guy et al., 2001). Male animals with a null allele of Mecp2 (NULL) develop severe neurological abnormalities and breathing irregularities (Viemari et al., 2005; Ogier et al., 2007; Roux et al., 2008; Zanella et al., 2008), and die between 8 and 12 weeks of life (Guy et al., 2001). Heterozygous female mice lacking one copy of Mecp2 develop similar phenotypes in a delayed and variable manner, likely due to variable patterns of X-chromosome inactivation in individual animals (Young and Zoghbi, 2004). Because this variability compounds experimental complexity, most research has been performed using male NULL mice.
Surprisingly, restoring MeCP2 function in all cells can reverse abnormalities in mice (Guy et al., 2007), demonstrating that loss of MeCP2 does not cause irreversible changes in cellular function and providing hope that the human disorder may be reversible. However, restoring function only to the forebrain did not rescue lethality (Alvarez-Saavedra et al., 2007), suggesting that the critical region for MeCP2 function is caudal to the forebrain.
One hypothesis is that specific clinical features present in RTT are the cumulative result of loss of MeCP2 function within subsets of cells. A variety of reports support the hypothesis that distinct phenotypes found in the NULL animals can be reproduced by removing MeCP2 in specific cell populations (Fyffe et al., 2008; Adachi et al., 2009; Samaco et al., 2009; Chao et al., 2010). Because the hindbrain includes neuronal circuits critical for the regulation of autonomic and respiratory function that may contribute to lethality, we hypothesized that the autonomic and respiratory dysfunction as well as the premature death seen in NULL animals is due to MeCP2 requirements within the hindbrain. To test this hypothesis, we either removed or restored MeCP2 function within the medulla and spinal cord. From this, we show that MeCP2 is both necessary and sufficient within this region for a number of physiological processes and normal lifespan.
Materials and Methods
Animals used in experiments.
All research and animal care procedures were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee and housed in the Association for Assessment and Accreditation of Laboratory Animal Care-approved animal facility at Baylor College of Medicine. Mecp2Tm1Bird mice (Guy et al., 2001) were obtained as a gift from Dr. Adrian Bird (University of Edinburgh, Edinburgh, UK) and backcrossed and maintained on a 129S6 background for >10 generations. HPRT-Cre mice (Su et al., 2002) were crossed to the Mecp2Tm1Bird mice to generate Mecp2Tm1.1Bird NULL mice. Mecp2Tm2Bird mice (Guy et al., 2007) were obtained from The Jackson Laboratory and maintained on a C57BL/6J background. Mecp2TM1Jae (Chen et al., 2001) animals were acquired from the University of California–Davis Mouse Mutant Resource Center and maintained on a C56BL/6J strain. HoxB1cre mice (Arenkiel et al., 2003) were maintained on a C57BL/6 background, and crossed to Mecp2Tm1Bird, Mecp2Tm2Bird, or Mecp2TM1Jae mice to generate male experimental animals. Identification of adult structures derived from the HoxB1 lineage was facilitated by using mice with an allele of nLacZ-IRES-myrGFP expressed under control of the Tau promoter, or an allele of LacZ integrated into the Rosa26 locus (Soriano, 1999; Hippenmeyer et al., 2005).
Western blot detection of proteins.
MeCP2 expression was determined by Western blot from brain tissue. Fresh brains were dissected from male mice and processed as hemibrains or further dissected to cerebrum and brainstem. The brains were homogenized in 100 mm Tris, pH 8.0, and 2% SDS via Dounce homogenizer with the B handle and rocked at room temperature for at least 1 h to ensure complete cell lysis. Soluble proteins were separated from cell debris by centrifugation at 20,000 × g for 15 min, and the supernatant was then loaded onto a PAGE gel and transferred onto a nitrocellulose membrane. The membrane was blocked in Tris-buffered saline, pH 7.6, and 0.1% Tween 20 (TBS-T) with 5% milk. For immunodetection of MeCP2, an antibody raised in rabbit (Shahbazian et al., 2002) diluted 1:2000 was used; detection of the loading control GAPDH was performed with an antibody raised in mouse (Millipore) diluted 1:20,000. HRP-conjugated secondary antibodies against mouse and rabbit were used at a dilution of 1:4000 (GE) and visualized with ECL detection reagent (GE), or Dura substrate (Pierce). The signals were captured on film and scanned or captured with a blot imager (Alpha Inotech) and quantified by densitometry with ImageJ software (Rasband, 1997–2011).
Glucose tolerance testing.
Male mice were fasted overnight for 16 h and then given an acute dose of glucose at 2 g (glucose)/kg (body weight) delivered as 300 mg/ml glucose prepared in normal saline. Blood glucose measurements were taken by tail bleeds of ∼10 μl per sample and analyzed with a OneTouch Ultramini blood glucometer. Samples were taken immediately before the glucose challenge and at 15, 30, 60, and 120 min after the glucose injection.
Chronic electroencephalographic recordings.
Male mice (aged 1.5 months) were anesthetized with Avertin (1.25% tribromoethanol/amyl alcohol solution, i.p.) using a dose of 0.02 ml/g. Teflon-coated silver wire electrodes (0.005 inch diameter) soldered to a microminiature connector were implanted bilaterally into the subdural space over frontal, central, parietal, and occipital cortices. Digital electroencephalographic (EEG) activity was monitored daily for up to 2 weeks during prolonged and random 2 h sample recordings (Stellate Systems, Harmonie software version 5.0b). A digital video camera was used to simultaneously monitor behavior during the EEG recording periods. All recordings were performed at least 24 h after surgery on mice freely moving in the test cage.
Heart rate data were collected by pulse oximetry with the Mouse Ox (Starr Life Sciences). Measurements were collected from either the tail or neck of male mice. For tail measurements, animals were placed within a restraint tube allowing access to the tail. For neck measurements, fur was removed around the neck with a depilatory cream. The mice were recorded for 5 min and their heart rate defined as the average over the 5 min recording. Heart rate data were collected 1–3 times per week throughout the lifespan of the animals. Analysis revealed stable heart rate throughout adulthood, aside from the time period 1 week preceding death, so the values were averaged for the individual animals over the course of their life, excluding the 7 d before the animal died spontaneously or became moribund and was killed, and then genotype averages were compared.
Measurement of body temperature.
Measurement of body temperature was performed in male mice with temperature monitoring implantable radio-frequency transponders (BMDS). The transponders were implanted subcutaneously such that the transponders would reside near the base of the neck. Temperature data were collected at least twice a week between 9:00 A.M. and 12:00 P.M. via a handheld transponder reader. Similar to the heart rate analysis, no change in temperature was observed over the lifespan of the animals except for the 7 d preceding death. Therefore, the values excluding the 7 d before death were averaged for each animal, and then genotype averages were compared.
Unrestrained whole-body plethysmography.
Male mice were placed within unrestrained whole-body plethysmography chambers (Buxco), ∼500 ml in volume with a continuous flow rate of 1 L/min flushing the chambers with fresh air. Mice were allowed to acclimate for 20 min, and baseline breathing was then recorded for 30 min. To determine response to hypoxic gas (10% O2, balance N2), the chamber was then flushed with hypoxic gas for 4 min and then breathing was recorded during the next 5 min of hypoxic exposure. Breath waveforms and derived parameters, including the instantaneous breathing rate, tidal volume, inspiratory time, and expiratory time, were identified and calculated with Biosystem XA software (Buxco). To reduce the artifacts from excessive movement and sniffing behavior, breaths that exhibited an inspiratory time <0.03 s, an expiratory time >10 s, and a calculated exhaled tidal volume >150% or <50% of calculated inhaled tidal volume were excluded; the recording was then split into 1 min intervals and only those minutes during which the animal spent <10% of its breaths above 500 breaths per minute were included in the analysis (results obtained by this method agreed closely with behavioral monitoring to exclude activity artifacts in a pilot experiment). Plethysmography data were collected in mice between 4 and 21 weeks of age with no significant effect of age on respiratory rate observed nor on tidal volume compared to age-matched controls with the exception of recordings performed immediately preceding death; therefore, values recorded within the last week of life were excluded. Breathing parameters for each animal during baseline and hypoxic challenge were determined as the average instantaneous value over the recorded interval and then averaged across trials. Respiratory parameters were then compared across genotypes.
Male mice were deeply anesthetized by intraperitoneal injection with Avertin at a dose of 0.04 ml/g and then fixed by transcardiac perfusion with PBS followed by 4% paraformaldehyde in PBS. Tissues for histological analysis were harvested: brains were excised from the skull, and the carotid body and superior cervical ganglion were dissected as preps attached to the carotid artery at the level of the carotid bifurcation into the internal and external carotid arteries. The tissues were then fixed overnight in 4% paraformaldehyde in PBS and cryopreserved by overnight incubations in increasing concentrations of sucrose (up to 30% sucrose). Tissues were embedded in O.C.T. compound (Sakura) and sectioned. For brain samples, 50 μm sections were collected with a cryostat and stained as floating sections. For carotid body and superior cervical ganglion samples, 25 μm sections were collected with a cryostat and stained as mounted sections on polylysine-coated slides. Sections were blocked for 1 h in a PBS solution containing 10% serum (matched to the host used for the secondary antibodies) and 0.3% Triton X-100. Primary and secondary antibody incubation was performed in the blocking solution for 48 h at 4°C for floating sections and 24 h at room temperature in a humidity chamber for slide-mounted sections. Sections were washed between incubations with PBS and 0.05% Triton X-100. If not included in the mounting medium, DAPI and Toto-3 were included in the penultimate wash. Primary antibodies were used at the following dilutions: rabbit anti-MeCP2 1:250 (Millipore), chicken anti-MeCP2 1:500 (Prosci), chicken anti-tyrosine hydroxylase 1:1000 (Abcam), chicken anti-β-galactosidase 1:1000 (Abcam), rabbit anti-NK1R 1:1000 (Sigma), rabbit anti-somatostatin 1:500 (Immunostar), goat anti-vesicular acetylcholine transporter 1:1000 (Millipore), and goat anti-Islet-1 1:500 (Neuromics). Secondary antibodies conjugated to DyLight 488, 549, or 649 were used at a dilution of 1:500 and raised in goat or donkey (Jackson ImmunoResearch). Sections were mounted with Prolong mounting medium (Invitrogen) and imaged via epifluorescent microscopy (Zeiss M1 with Axiovision software) or laser scanning confocal imaging (Zeiss 710 with ZEN software). Stereological quantification of cells expressing MeCP2 in the hindbrain was performed on coronal sections through the pons and medulla with systematic random sampling of every 12th section, with optical fractionators measuring 144.72 μm × 144.72 μm set at an area sampling fraction of 1/64 through the full thickness of the tissue excluding the top plane of focus (West et al., 1991). The pons was defined as ventral to the aqueduct in sections caudal to the substantia nigra through sections containing the locus ceruleus (LC). The medulla was defined as the first section caudal to the locus ceruleus through to the beginning of the spinal cord. Quantification of tyrosine hydroxylase (TH)-positive cells expressing MeCP2 was performed by examining the full cross-sectional area of every eighth section, excluding cells present in the top plane of focus to avoid overestimation of cell numbers, and first identifying cells expressing TH followed by checking for MeCP2 expression. TH-expressing cells were divided into A1, A2, A5, and LC populations based on anatomical location (Paxinos and Franklin, 2001). The nucleus of the solitary tract (NTS) was identified by somatostatin expression in the dorsal medulla (Finley et al., 1981); MeCP2-expressing cells were estimated using the optical fractionator technique sampling every 12th slide with a sampling grid area of 22,500 μm2, dissector height of 15 μm, and dissector area of 900 μm2. The retrotrapezoidal nucleus (RTN) was defined as the region ventral to staining for VAchT or ISL1 as a marker of the facial nucleus (Lazarenko et al., 2009); MeCP2-expressing cells were estimated using the optical fractionator technique sampling every fourth slide with a sampling grid area of 22,500 μm2, dissector height of 15 μm, and dissector area of 5625 μm2. The pre-Botzinger was defined as the somatostatin- or NK1R-stained region ventral to staining for VAchT or ISL1 as a marker of the nucleus ambiguus (Stornetta et al., 2003; Rose et al., 2009); MeCP2-expressing cells were estimated using the optical fractionator technique sampling every fourth slide with a sampling grid area of 8100 μm2, dissector height of 15 μm, and dissector area of 2025 μm2. The carotid body and superior cervical ganglion were identified by the presence of TH-stained cells at the bifurcation of the carotid artery, and rostral and deep to the carotid artery, respectively (Roux et al., 2008). Quantification was performed using ImageJ with the LOCI Bioformats plug-in (Rasband, 1997–2011) or Stereo Investigator software (MBF Bioscience).
Motor tasks (dowel walking, wire hang, footslip, accelerating rotarod), social interaction (partition test), learning (conditioned fear), anxiety (elevated plus maze, light–dark), and open-field analysis were performed on male mice as previously described (Samaco et al., 2008).
All statistics were performed using SPSS v.12 on a PC. Parametric statistics were performed using ANOVA with genotype as a factor. For conditions in which the genotype factor had more than two levels, formalized post hoc testing was performed using Tukey's HSD correction to detect pairwise differences. Survival analysis was performed using Kaplan–Meier survival analysis, with Tarone–Ware method applied to detect differences in survival between genotype groups.
Removal of MeCP2 from the HoxB1 domain targets the hindbrain and several key respiratory and autonomic centers
To remove MeCP2 from the medulla, we bred mice that contain a conditional Mecp2 allele flanked by loxP sequences (Mecp2flox/+) (Guy et al., 2001) to mice that express the Cre recombinase in the HoxB1 domain (HoxB1cre) (Arenkiel et al., 2003). HoxB1cre drives Cre expression in the rhombencephalon distal to the posterior border of rhombomere 3, which encompasses the caudal part of the pons, the medulla, and the spinal cord. To verify that MeCP2 protein expression was disrupted specifically within the hindbrain of HoxB1cre/WT; Mecp2flox/Y (hereafter referred to as CKO for “conditional knock-out”) animals, we performed Western blotting on both the hindbrain and forebrain tissues (Fig. 1A), including samples from other genotypes (HoxB1WT/WT; Mecp2WT/Y referred to as “WT,” HoxB1cre/WT; Mecp2WT/Y referred to as “CRE,” and HoxB1WT/WT; Mecp2flox/Y referred to as “FLOX”) as littermate controls. MeCP2 protein was absent from the hindbrain of CKO animals but present in the forebrain. In contrast, MeCP2 was present in both brain regions of the control genotypes. The ratio of MeCP2 in the hindbrain to that in the forebrain was dramatically reduced in the CKO animals compared to all other genotypes (Fig. 1B).
To further evaluate the regions targeted by the HoxB1cre, we bred the HoxB1cre mice with reporter mice either possessing an allele of nLacZ-IRES-myrGFP expressed under control of the Tau promoter or a conditional β-galactosidase reporter integrated into the ROSA26 locus (Soriano, 1999; Hippenmeyer et al., 2005). Both of these reporter alleles express β-galactosidase in a Cre-dependent fashion, one in all tissues and one only in neurons. Within the brain, the rostral boundary of the HoxB1 domain lies within the pons (Fig. 1C). Due to curvature of the neuraxis, the dorsal pons represents more rostrally defined structures and is excluded from the HoxB1 domain (Fig. 1D). However, the ventral pons is derived from a more caudal origin and is within the HoxB1 domain. In contrast, the medulla was completely derived from the HoxB1 domain (Fig. 1E). Within the hindbrain, we investigated several subpopulations of neurons involved in autonomic and breathing control to ensure that they were targeted by HoxB1cre. The A5, A2, and A1 noradrenergic neurons are all targeted by HoxB1cre (Fig. 1F,H,I), while the LC in the dorsal pons is not (Fig. 1G). Key respiratory centers of the ventral medulla such as the retrotrapezoid nucleus and pre-Botzinger complex are also included within the HoxB1 domain (Fig. 1J,K). The peripherally located carotid body, which is critical for recognizing blood oxygenation to modulate respiratory function, is also targeted by HoxB1cre (Fig. 1L,M). Together, these data show that the HoxB1cre targets several regions critical for regulation of respiratory and autonomic function, but excludes some such as the locus ceruleus.
Removing MeCP2 from the HoxB1 domain impairs motor coordination but does not modify forebrain-associated behaviors
Male mice with a null allele of Mecp2 exhibit a variety of movement problems (Chen et al., 2001; Guy et al., 2001; Pratte et al., 2011). Our CKO animals performed poorly on a number of motor coordination tasks compared to control genotypes, including dowel rod walking (Fig. 2A) and wire hanging (Fig. 2B). Moreover, CKO animals were not able to walk well on a metal grid, having more footslips than all control genotypes (Fig. 2C). CKO animals also performed poorly on the accelerating rotating rod (Fig. 2D).
In addition, we performed a large battery of behavior tests on the CKO animals. The CKO animals did not show any abnormalities in the open-field analysis (Fig. 2E), and in a test of social interaction (the partition test), the CKO animals showed no difference compared with the FLOX animals (Samaco et al., 2008) (Fig. 2F). No difference was observed in the light–dark task, a test of anxiety (Fig. 2G). In a test of learning and memory, the conditioned fear task, the CKO animals showed the same degree of learning as the FLOX allele (Samaco et al., 2008) (Fig. 2H). Therefore, although MeCP2 function is required within the HoxB1 domain for motor coordination, it is not required within this region for general activity, social function, anxiety, or fear-based learning.
Animals lacking MeCP2 in the HoxB1 domain have normal cortical EEG patterns
Because seizures are common in RTT (Glaze et al., 2010), we evaluated these mice for cortical electrical abnormalities. NULL animals also show abnormal cortical EEG tracings, with a variety of features that indicate increased excitability. Figure 2I shows a representative epoch from one animal with abnormal synchronous cortical discharge activity. At other times, brief spike bursts and a recorded seizure characterized by an abrupt onset of generalized spike discharges that gradually increased in frequency until the termination of the ictal event were recorded. In contrast, the EEG pattern of the CKO animals looked remarkably normal (representative trace in Fig. 2J), and no overt seizures were ever observed. These data indicate that the abnormal cortical electrical activity observed in the NULL animals is due to loss of MeCP2 function in the non-HoxB1 domain, likely within the cortex itself.
Removing MeCP2 from the HoxB1 domain disrupts a subset of physiological processes
Because genetic strain effects can dramatically alter phenotypes produced when MeCP2 function is disrupted, we performed detailed characterization of Mecp2tm1.1Bird/Y animals (hereafter referred to as “NULL”) (Guy et al., 2001) in the same genetic strain background as the CKO animals. In this background, the NULL animals are markedly obese (Fig. 3A). In contrast, the CKO animals are not overweight and, in fact, weigh less than FLOX littermate control animals from 9 to 10 weeks of life (Fig. 3B). NULL mice show an abnormal response to injected glucose with increased serum glucose at all time points after injection (Fig. 3C). In contrast, the CKO show normal responses to this glucose challenge (Fig. 3D).
To determine whether the animal models of RTT have cardiac rhythm abnormalities, we characterized the NULL animals with a noninvasive measure of heart rate. NULL animals have decreased basal heart rate (Fig. 3E), and this finding is reproduced in CKO animals (Fig. 3F). To further characterize autonomic dysfunction, we measured body temperature with an implanted temperature probe. NULL animals show a decreased basal temperature (Fig. 3G), which is not present in the CKO animals (Fig. 3H). Thus, MeCP2 function is required within the hindbrain for normal heart rate, but is dispensable in this region for body temperature. We have previously found that MeCP2 is required within the hypothalamus for normal weight regulation. Because the hypothalamus is rostral to the boundary of the HoxB1 domain targeted in CKO animals (Fig. 1C), the lack of obesity in CKO animals was expected; however, their underweight phenotype implies an additional role of MeCP2 in the HoxB1 domain for weight regulation. The normal temperature regulation found in these CKO animals suggests that MeCP2 function is required in non-HoxB1 lineages for correct temperature homeostasis. Blood oxygenation, as measured by pulse oximeter, was unaffected both in NULL animals (Fig. 3I) and in CKO mice (Fig. 3J).
Loss of MeCP2 from the HoxB1 domain caused early lethality
NULL animals died early, with a median survival of 65 d (Fig. 4A). Animals that lack MeCP2 in the HoxB1 region showed similar early lethality, with a median survival of 93 d (Fig. 4A). Additionally, we found similar early death when we removed MeCP2 from the HoxB1 domain using a different “floxed” allele of Mecp2, Mecp2tm1Jae (median age of death = 126 d, n = 13) (Chen et al., 2001; Guy et al., 2001; Pratte et al., 2011). We noted that NULL animals had a progressive decline in their heart rate as they approached death (Fig. 4B) and a decline in their basal temperature on the day of death (Fig. 4C), with no evidence of infection as assessed via blood cultures. Additionally, continuous EEG recorded from a subset of moribund NULL animals did not show increased epileptiform or seizure activity. Finally, telemetric recording of cardiac electrical activity in moribund animals showed progressive bradycardia, and in a few cases asystole, but no evidence of other cardiac rhythm abnormalities that could cause death. We noted a similar decline in heart rate (Fig. 4B) and body temperature (Fig. 4C) in the CKO animals on the day of death. Therefore, removing MeCP2 from the hindbrain caused death similar to that observed in NULL animals, indicating that MeCP2 is necessary within the HoxB1 domain for normal lifespan.
Animals lacking MeCP2 in the HoxB1 domain exhibited abnormal breathing responses
Abnormal patterns of breathing are present in children with MECP2 mutations and have been reported in NULL mice (Viemari et al., 2005; Ogier et al., 2007; Voituron et al., 2009). To accurately compare our CKO animals to NULL animals, we performed detailed characterizations in the same genetic strain. In this strain, the NULL animals had an increased mean breathing rate compared with WT controls (Fig. 5A). This increased breathing rate was not accompanied by an alteration in tidal volume unless normalized for body weight (Fig. 5B,C); however, the obesity phenotype of the NULL animals likely drove this difference in the weight normalized tidal volume. Examination of the average inspiratory and expiratory times revealed that a decrease in the expiratory time drove the increased baseline breathing rate observed in NULL animals (Fig. 5D,E). In contrast, the hindbrain CKO animals did not recapitulate this increased baseline breathing rate (Fig. 5F). Similarly, the CKO did not exhibit an altered tidal volume unless normalized for body weight (Fig. 5G,H); however, the underweight phenotype of the CKO animals likely contributed to this phenotype. The CKO exhibited a baseline inspiratory time between that of the FLOX animals and the WT and CRE animals, although not significantly different from either (Fig. 5I). Also, although the expiratory time of the CKO animals was significantly decreased relative to WT and CRE animals, it was not different from FLOX littermates and not sufficient to recapitulate the increased baseline breathing rate observed in the NULL animals (Fig. 5J).
NULL animals have also been reported to show abnormal breathing responses to challenges with decreased oxygen concentrations (10% O2, balance N2—hypoxia) (Roux et al., 2008; Voituron et al., 2009). We found that NULL animals showed an increased respiratory rate relative to WT littermates in response to hypoxia (Fig. 5A), indicative of a deficit in the hypoxic ventilatory decline associated with exposure to acute hypoxia (Teppema and Dahan, 2010). The tidal volume of NULL animals during hypoxia was similar to WT littermates; however, the obesity phenotype of the NULL animals contributes to a decreased weight normalized tidal volume during hypoxia (Fig. 5B,C). Assessment of the average inspiratory and expiratory times revealed that a decrease in the expiratory time also drove the increased hypoxic breathing rate observed in NULL animals (Fig. 5D,E). Removing MeCP2 from the HoxB1 domain recapitulated this increased hypoxic breathing rate relative to control littermates (Fig. 5F). The tidal volume of CKO animals was not significantly different from WT littermates, although FLOX animals did exhibit an increased tidal volume versus the CKO (Fig. 5G). Furthermore, there were no significant differences between CKO and control littermates for tidal volume normalized to bodyweight during hypoxia. The inspiratory time of FLOX and CKO animals was slightly decreased versus CRE littermates (Fig. 5I); however, the decreased expiratory time of the CKO animals versus control littermates during hypoxia provided a larger contribution to the increased breathing rate during hypoxia (Fig. 5J).
Together these data suggest that the increased respiratory rates resulting from MeCP2 dysfunction during baseline conditions or challenges with hypoxia possess different origins. Baseline abnormalities likely originate from structures rostral to the HoxB1 domain and abnormalities during hypoxia from structures within the HoxB1 domain.
Restoring MeCP2 function within the HoxB1 domain rescued autonomic abnormalities and lifespan
To determine whether MeCP2 expression in this tissue is sufficient to prevent early death and autonomic abnormalities, we restored MeCP2 exclusively within the HoxB1 domain using a recently engineered stop-flox or “RESCUE” allele of Mecp2, Mecp2tm2Bird (Guy et al., 2007). In animals lacking Cre-recombinase, no Mecp2 mRNA or MeCP2 protein is expressed (NON-RESCUE), and thus these animals are indistinguishable from NULL mice. MeCP2 expression can be induced with exposure to Cre, such as the HoxB1cre (RESCUE). Western blotting showed that MeCP2 expression was absent from both the forebrain and the hindbrain of NON-RESCUE animals, whereas RESCUE animals showed MeCP2 expression within the hindbrain but not the forebrain (Fig. 6A). The amount of MeCP2 protein induced within the hindbrain was <50% of that seen in WT control animals (Fig. 6B). Stereological quantification of MeCP2-expressing cells throughout the pons and medulla revealed that RESCUE animals possess fewer MeCP2-expressing cells in the hindbrain than WT littermates (WT, n = 3, pons 1.65 × 106 ± 0.44 × 106, medulla 1.01 × 106 ± 0.08 × 106; RESCUE, n = 3, pons 0.39 × 106 ± 0.07 × 106, medulla 0.49 × 106 ± 0.08 × 106; AVG ± SD) (Fig. 6D).
RESCUE animals possessed regional restriction of MeCP2-expressing cells expected from the HoxB1cre (Fig. 6C,E–H—compare to Fig. 1C–E). To determine whether the efficiency of MeCP2 expression in RESCUE animals reflected a general phenomenon or biased restoration to specific populations of neurons, we examined the efficiency of MeCP2 expression in several regions implicated in respiratory and autonomic control. We quantified the number of noradrenergic neurons and the fraction that expressed MeCP2 in WT and RESCUE animals (Fig. 6I–L″). WT animals expressed detectable MeCP2 in nearly 100% of the noradrenergic neurons quantified [WT; A1: 1011 ± 101 TH(+) cells, 92.5 ± 6.4% TH(+)MeCP2(+); A2: 1435 ± 264 TH(+) cells, 97.8 ± 0.5% TH(+)MeCP2(+); A5: 477 ± 76 TH(+) cells, 90.7 ± 13.7% TH(+)MeCP2(+); LC: 3043 ± 434 TH(+) cells, 99.1 ± 0.8% TH(+)MeCP2(+); n = 3; AVG ± SD]; rescue animals expressed detectable MeCP2 in approximately half of the noradrenergic neurons of the A1, A2, and A5 populations and practically none of the LC neurons, which lie outside of the HoxB1 domain [RESCUE; A1: 1332 ± 154 TH(+) cells, 52.0 ± 3.8% TH(+)MeCP2(+); A2: 1844 ± 237 TH(+) cells, 42.1 ± 4.2% TH(+)MeCP2(+); A5: 760 ± 513 TH(+) cells, 56.1 ± 10.2% TH(+)MeCP2(+); LC: 2475 ± 538 TH(+) cells, 7.3 ± 7.1% TH(+)MeCP2(+); n = 3; AVG ± SD]; and NON-RESCUE animals expressed detectable MeCP2 in none of the noradrenergic populations tested [NON-RESCUE; A1: 1533 ± 300 TH(+) cells, 0 ± 0% TH(+)MeCP2(+); A2: 1883 ± 159 TH(+) cells, 0 ± 0% TH(+)MeCP2(+); A5: 680 ± 233 TH(+) cells, 0 ± 0% TH(+)MeCP2(+); LC: 2795 ± 484 TH(+) cells, 0 ± 0% TH(+)MeCP2(+); n = 3; AVG ± SD]. To estimate the efficiency of MeCP2 expression in the NTS, the major relay center for many autonomic and respiratory reflex functions, we quantified the MeCP2-expressing cells contained within somatostatin-stained regions of the dorsal medulla (Fig. 6M,M′,M″) [WT 87,476 ± 17,332 MeCP2(+) cells; RESCUE 32,382 ± 6362 MeCP2(+) cells; NON-RESCUE 0 ± 0 MeCP2(+) cells; n = 3 per genotype; AVG ± SD]. To estimate the efficiency of MeCP2 expression in the RTN, a major respiratory control center, we quantified MeCP2-expressing cells ventral to the facial nucleus (Fig. 6N,N′,N″) [WT 6725 ± 1815 MeCP2(+) cells; RESCUE 2740 ± 719 MeCP2(+) cells; NON-RESCUE 0 ± 0 MeCP2(+) cells; n = 3 per genotype; AVG ± SD]. To estimate the efficiency of MeCP2 expression in the pre-Botzinger complex, the major respiratory rhythm generator, we quantified the MeCP2-expressing cells in the somatostatin and NK1R-expressing region ventral to the nucleus ambiguus (Fig. 6O,O′,O″) [WT 10,201 ± 608 MeCP2(+) cells; RESCUE 4400 ± 1092 MeCP2(+) cells; NON-RESCUE 0 ± 0 MeCP2(+) cells; n = 3 per genotype; AVG ± SD]. Furthermore, RESCUE animals also restored MeCP2 expression to peripheral regions, including the superior cervical ganglion (Fig. 6P,P′,P″) and the carotid body (Fig. 6Q,Q′,Q″).
The NON-RESCUE animals showed early lethality similar to that observed in the NULL animals, with a median survival of 95 d (Fig. 7A). In contrast, despite our observation that only a fraction of cells in the hindbrain have been induced to express MeCP2 (Fig. 6D), RESCUE animals showed a substantial improvement in survival (p < 0.001) (Fig. 7A). Only five animals died or became moribund and were killed (out of an initial cohort of 16 animals), with the remainder of the animals surviving >45 weeks, at which point some animals were killed for tissue collection.
In addition to rescuing the early lethality, restoring MeCP2 in this region also significantly reversed other physiological changes. In this genetic background, the NON-RESCUE animals were slightly underweight, whereas restoring MeCP2 function caused the RESCUE animals to become markedly obese (Fig. 7B), a phenotype consistent with the lack of MeCP2 within the forebrain (Fyffe et al., 2008). The NON-RESCUE animals had the same decreased heart rate seen in the NULL animals (Fig. 7C), whereas the RESCUE animals had a heart rate rescued to a WT level. However, no apparent differences in blood oxygenation existed between the genotypes (Fig. 7D).
Mimicking the NULL animals (Fig. 5A), both the NON-RESCUE and the RESCUE animals showed increased baseline breathing (Fig. 5K). Thus, restoring MeCP2 function within the hindbrain did not rescue this baseline breathing abnormality. In contrast, whereas the NON-RESCUE animals showed the same relative increased breathing rate in response to hypoxia (Fig. 5K) seen in the NULL animals (Fig. 5A), the RESCUE animals showed a rescue of the breathing rate during hypoxia similar to that in WT controls (Fig. 5K). Additionally, tidal volume was generally unaffected by genotype during baseline unless normalized to body weight, in which case body weight phenotypes had a large impact on the normalized values (Fig. 5L,M). Similarly, during hypoxia, the tidal volume of RESCUE animals was not significantly different from WT animals, although NON-RESCUE animals possessed a tidal volume slightly lower than WT littermates; normalization to body weight caused the RESCUE animals to exhibit a decreased normalized tidal volume relative to littermates largely impacted by the obesity of the animals (Fig. 5L,M). Inspiratory time was unaffected by genotype during either baseline or hypoxic conditions. In contrast, the increased breathing rate phenotypes were driven by decreased expiratory times similar to the effects seen in NULL and CKO animals.
Cumulatively, these results indicate that MeCP2 function within a fraction of cells in the HoxB1 lineage is indeed sufficient to restore normal lifespan and to rescue some of the autonomic abnormalities seen in the NULL animals. Importantly, the continued presence of some observed physiological abnormalities such as abnormal baseline breathing in animals with improved survival indicates that these abnormalities are not the sole and direct cause of death.
Traditionally, understanding the role of a specific gene product in a particular cell population, or region, of the brain relies on cell- or tissue-specific deletion of the gene through a conditional knock-out approach. This conditional approach has been successfully used to understand the cell-autonomous role of MeCP2 (Fyffe et al., 2008; Adachi et al., 2009; Samaco et al., 2009; Chao et al., 2010) and has been invaluable toward determining the requirement for a gene's function in a specific cell population or region. However, in the context of distributed neural circuits coordinating behavioral and physiological processes, removing a gene from any single component of the circuit may completely disrupt the overall circuit's function. There may be many possible ways to disrupt the function of the circuit by removing the gene of interest from any number of cellular populations within that circuit. On the other hand, the ability to rescue a particular phenotype by restoring MeCP2 function within a select region or group of cells demonstrates conclusively that the entire neural circuit function has been restored. In the case of MeCP2 dysfunction, this has important clinical implications: the identification of an anatomically restricted region in which restoration of MeCP2 function causes dramatic changes in clinically relevant phenotypes implies that specific therapy, such as gene therapy, targeted toward this restricted anatomic area may have dramatic effects on clinical features.
MeCP2 function is required in the anatomical region defined by the HoxB1 lineage for normal lifespan. Furthermore, restoring MeCP2 function within this same region restores lifespan, demonstrating that MeCP2 function is both necessary and sufficient within the HoxB1 lineage for survival. In addition to the effect on lifespan, several autonomic and respiratory phenotypes are dependent on MeCP2 function within the HoxB1 domain. In particular, loss of MeCP2 in this region contributes to a decreased heart rate as well as a greater increase in respiratory rate during an acute hypoxic challenge. There is clear complementarity in these phenotypes, as they result from loss of MeCP2 function within the HoxB1 domain and are reversed when MeCP2 function is restored within this domain (Table 1). Survival, weight, heart rate, basal hyperventilation, and hypoxic respiratory response all exhibited reciprocal phenotypes between the HoxB1 CKO and RESCUE models. The ability to identify such reciprocal phenotypes between the CKO and RESCUE animals argues strongly for a tissue-autonomous function for MeCP2 in the hindbrain. Furthermore, the reciprocity of the weight phenotypes observed between the CKO and RESCUE mice (loss of MeCP2 within the HoxB1 domain causing an underweight phenotype, and loss of MeCP2 rostral to the HoxB1 domain causing an overweight phenotype) revealed that MeCP2 is required within two separate anatomical domains for competing drives affecting weight.
Interestingly, some of the abnormalities observed in the NULL mice are not reproduced when MeCP2 function is removed from the HoxB1 domain and are not rescued when MeCP2 expression is restored in this domain (Table 1). This indicates that such phenotypes are clearly the result of MeCP2 dysfunction in the region outside the HoxB1 domain. Of the phenotypes not dependent on MeCP2 function within the HoxB1 domain, perhaps the most notable is the increased breathing rate observed in the NULL animals during baseline breathing. This breathing abnormality is not present in CKO animals but is present in the RESCUE animals, suggesting that the phenotype has an origin outside the HoxB1 expression domain. The fact that some of the breathing abnormalities are dependent on MeCP2 function within the HoxB1 domain (i.e., response to hypoxia), whereas other breathing abnormalities, such as the basal hyperventilation, are not, demonstrates that MeCP2 function is important in multiple nodes of respiratory circuitry, and that the function within these anatomically distinct regions are separable. Furthermore, the increase in breathing rate, during basal conditions or during hypoxic challenge, is largely due to a reduction of the expiratory time per breath, a result that is consistent with models of increased net activation of respiratory pattern generating preinspiratory neurons of the pre-Botzinger complex (Smith et al., 2009). Given the anatomical separation in the requirement of MeCP2 for normal baseline or hypoxic breathing rates, the requirement of MeCP2 rostral to the HoxB1 domain for normal baseline respiration suggests impaired signaling from nuclei rostral to the HoxB1 domain, possibly including the parabrachial nuclei and the Kölliker–Fuse nucleus, to the ventral respiratory group of the medulla (Stettner et al., 2007; Song et al., 2010). In contrast, the requirement within the HoxB1 domain for normal respiratory rate during hypoxia suggests impairments in the circuits responding to oxygen levels. One possibility is that signaling from the NTS relaying the sensation of hypoxia detected in the carotid body to the ventral respiratory group of the medulla is altered (Teppema and Dahan, 2010); specifically, deficits in GABA signaling caused by loss of MeCP2 (Medrihan et al., 2008; Chao et al., 2010) may underlie the apparently impaired hypoxic ventilatory decline caused by loss of MeCP2 in the hindbrain, presenting as relative hyperventilation during an acute hypoxic challenge (Teppema and Dahan, 2010). Alternatively, because the carotid body itself is targeted in the CKO and RESCUE animals by HoxB1Cre, direct involvement of this oxygen sensor is also a possibility.
Additionally, the origin of the abnormal response to hypoxia within the HoxB1 domain is particularly important because it has been proposed that early exposure to hypoxic stress may be an inciting feature in the development of the disrupted breathing in the mouse models as well as in girls with RTT (Voituron et al., 2009). However, the occurrence of the abnormal basal breathing rate in the RESCUE animals despite a normalized respiratory rate during hypoxia suggests that basal hyperventilation is not secondary to the chemosensory deficits. It remains to be seen whether the chemosensory deficits precede other respiratory deficits not explored in this study, or more specifically whether exposure to hypoxic stress plays a causal role in the progression of other autonomic deficits.
A surprising finding of this study is that restoring MeCP2 protein to only 50% of the cells within the hindbrain results in rescue of lifespan and other autonomic and breathing abnormalities. Detailed stereological quantification of MeCP2 expression across several regions of the hindbrain suggests that the approximate 50% efficiency of rescue was uniform throughout the HoxB1 domain. Similar phenotypic and histological characterization using more regionally restricted Cre lines that subdivide the HoxB1 domain are likely to refine our understanding of the anatomical requirements of MeCP2 function. However, the apparent uniformity of the modest efficiency of rescue suggests that restoring MeCP2 function in some cells has non-cell-autonomous effects. Restoring MeCP2 may result in production of secreted factors such as growth factors that improve the function of nearby cells. On the other hand, the effect may be at the circuit level, where only a fraction of the cells within a circuit that express MeCP2 are capable of restoring the entire circuit's function. These possibilities are not mutually exclusive, and the overall positive effect of restoring MeCP2 function in a fraction of cells in a region may be due to both possibilities. However, there is a requirement to reestablish MeCP2 in particular anatomic regions to induce such a dramatic effect, because restoring MeCP2 function solely within the forebrain does not rescue lifespan (Alvarez-Saavedra et al., 2007). In the context of developing therapies to treat RTT, aside from the technical challenge of effectively restoring MeCP2 function in specific cells within the human brain, an additional challenge will be to determine exactly which cellular populations have critical requirements for MeCP2 function to modulate specific phenotypes. The work outlined here provides a methodology for identifying these cellular populations, which then could be useful in the development of novel regionally targeted therapies.
This work was supported by the Cynthia and Anthony Petrello Scholar fund at the Jan and Dan Duncan Neurological Research Institute, Texas Children's Hospital (J. L. Neul), U.S. National Institutes of Health Grants NS52240 (J.L. Neul), HD062553 (J.L. Neul), NS066601 (C.S.W.), NS29709 (J. L. Noebels), and HD24064 (Baylor Intellectual and Developmental Disabilities Research Center), and the International Rett Syndrome Foundation. We thank Huda Zoghbi, Rodney Samaco, Adriano Flora, Alexander Bassuk, Stephen Maricich, Tiemo Klisch, John Fryer, and Benjamin Arenkiel for useful scientific discussions and advice. We thank Sean Cullen, Michael McQueen, Sarah-Ann Pike, and Ekaette Shammah for technical assistance, Corinne Spencer and Richard Paylor and the Neurobehavioral Core for advice on behavioral experiments, Corry Reynolds and the Baylor Mouse Phenotyping Core for assistance and equipment for the plethysmography experiments, and Hui-Chen Lu for access to equipment for stereology experiments. The project described was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (EKS NICHD). The content is solely the responsibility of the authors and does not necessarily represent the official views of the EKS NICHD or the National Institutes of Health.
The authors declare no competing financial interests.
- Correspondence should be addressed to Jeffrey L. Neul, Jan and Dan Duncan Neurological Research Institute, 1250 Moursund Street, Suite 1250, Houston, TX 77030.